Scholarlink

<p>Catastrophic rockslides seriously threaten human life and property in river valleys. Quantitative risk assessment (QRA) of rockslides requires a reliable evaluation of failure probabilities, which hinges on addressing three critical challenges: selecting an appropriate stability evaluation model, managing input parameter uncertainties, and accurately identifying potential slip surfaces. This study proposes an innovative framework that integrates Bayesian back analysis and the three-dimensional limit equilibrium method (3D LEM) to enhance the reliability of failure probability predictions for rock slopes. The framework incorporates a sloping local base level method for slip surface identification, calibrated sequentially for potentially unstable rock slopes. A MATLAB implementation of 3D LEM, accounting for dual-layer lithological compositions and irregular slip surfaces, was developed to calculate the factor of safety. To efficiently estimate failure probabilities, a Kriging surrogate model was combined with Monte Carlo simulations. The framework was validated using the two consecutive slides of the Baige rockslide in 2018. Bayesian back analysis refined the shear strength parameters based on observed stability conditions of the first slide. The posterior distribution derived from this analysis predicted a failure probability of 94.06% for the second slide, aligning closely with observed outcomes. Subsequently, the posterior distribution from the second analysis was applied to assess failure probabilities in other potentially unstable slopes. The results demonstrate the proposed framework’s robustness and accuracy, offering a reliable tool for assessing rock slope stability. This study provides valuable insights for advancing QRA methodologies, aiding the mitigation of rockslide risks in river valley regions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

3D probabilistic assessment of potentially unstable rock slope failure via sequential Bayesian updating

  • 期刊论文

  • 发表日期：

    2025年3月31日

查看全文
  • Sheng Wang,
  • Peng Zeng,
  • Tianbin Li,
  • Xiaoping Sun,
  • Caichu Xia,
  • Binyu Xu

摘要

Catastrophic rockslides seriously threaten human life and property in river valleys. Quantitative risk assessment (QRA) of rockslides requires a reliable evaluation of failure probabilities, which hinges on addressing three critical challenges: selecting an appropriate stability evaluation model, managing input parameter uncertainties, and accurately identifying potential slip surfaces. This study proposes an innovative framework that integrates Bayesian back analysis and the three-dimensional limit equilibrium method (3D LEM) to enhance the reliability of failure probability predictions for rock slopes. The framework incorporates a sloping local base level method for slip surface identification, calibrated sequentially for potentially unstable rock slopes. A MATLAB implementation of 3D LEM, accounting for dual-layer lithological compositions and irregular slip surfaces, was developed to calculate the factor of safety. To efficiently estimate failure probabilities, a Kriging surrogate model was combined with Monte Carlo simulations. The framework was validated using the two consecutive slides of the Baige rockslide in 2018. Bayesian back analysis refined the shear strength parameters based on observed stability conditions of the first slide. The posterior distribution derived from this analysis predicted a failure probability of 94.06% for the second slide, aligning closely with observed outcomes. Subsequently, the posterior distribution from the second analysis was applied to assess failure probabilities in other potentially unstable slopes. The results demonstrate the proposed framework’s robustness and accuracy, offering a reliable tool for assessing rock slope stability. This study provides valuable insights for advancing QRA methodologies, aiding the mitigation of rockslide risks in river valley regions.